Skills
97,222 packaged capabilities and playbooks for agents. Install any of them with one command.
Agent workflows
Quality gate via second model. Spawn a different AI model to review work before committing. Includes refusal routing: if one model refuses, switch silently to the next. Extended in v0.25.1 with structured review-mode gating (when to invoke vs not) and a Codex code-review handoff for the diff-review case
Creates a comprehensive, context-rich implementation plan through deep codebase analysis, a short clarifying interview, and external research. Accepts a tracker ticket (a Jira/Linear/GitHub key or URL, fetched from the tracker) or a free-form feature request. Use when you have a ticket or feature and need a one-pass-ready plan before writing any code.
Enterprise AI adoption platform for Claude Code — measures team usage via OpenTelemetry, syncs skills/MCP servers/hooks from a central dashboard, and delivers context-aware skill suggestions at prompt time. Solves the Intention-Action Gap: organizations that deploy Zeude see 3x adoption improvement (6% → 18%). Requires Supabase (config) and ClickHouse (analytics). Triggers on: zeude, ai adoption, claude code adoption, enterprise claude, opentelemetry claude, skill sync, team claude management.
Analyze the current conversation's tool calls, responses, errors, and approach to produce a structured feedback report for improving MCP tool usage with ARC-1. Use when the user asks to "analyze this session", "review my chat", "give feedback on tool usage", "how did I do", or wants a retrospective on a debugging session.
Delegate coding tasks to Codex CLI for execution, or discuss implementation approaches with it. CodeX is a cost-effective, strong coder — great for batch refactoring, code generation, multi-file changes, test writing, and multi-turn implementation tasks. Use when the plan is clear and needs hands-on coding. Claude handles architecture, strategy, copywriting, and ambiguous problems better.
Unified Minions skill for both deterministic shell jobs and LLM subagent orchestration. Replaces the older `gbrain-jobs` routing intent. Use when: submitting gbrain jobs, shell/background tasks, spawning subagents, checking progress, steering running work, pausing/resuming, parallel fan-out. One durable, observable, steerable queue interface. Also carries the durable-execution doctrine for any operation expected to exceed ~2 minutes: capability ladder, deadman checks that verify the result was reported, and content-addressed stage checkpoints for expensive pipelines
Creates comprehensive handoff documents for seamless AI agent session transfers. Triggered when: (1) user requests handoff/memory/context save, (2) context window approaches capacity, (3) major task milestone completed, (4) work session ending, (5) user says 'save state', 'create handoff', 'I need to pause', 'context is getting full', (6) resuming work with 'load handoff', 'resume from', 'continue where we left off'. Proactively suggests handoffs after substantial work (multiple file edits, complex debugging, architecture decisions). Solves long-running agent context exhaustion by enabling fresh agents to continue with zero ambiguity.
Analyze your SpecStory AI coding sessions in .specstory/history for yak shaving — when your initial goal got derailed into rabbit holes. Run when user says “analyze my yak shaving”, “check for rabbit holes”, “how distracted was I”, or “yak shave score”.
Build tools that agents can use effectively, including architectural reduction patterns. Use when creating new tools for agent systems, debugging tool-related failures or misuse, or optimizing existing tool sets for better agent performance.
Build AI agents with the Strands Agents SDK—the open-source framework (the agent “brain”) for writing agent logic, tools, and multi-agent systems in Python. Model-agnostic, with AWS Bedrock by default. Covers Agent, the @tool decorator, model providers (BedrockModel), multi-agent patterns (agents-as-tools, Swarm, Graph), conversation management, and streaming. Every import is verified against official Strands docs. To deploy a Strands agent on AWS, use the aws-harness skill. Triggers on Strands, Strands Agents, Strands SDK, agent framework, agents as tools, Swarm, Graph multi-agent, BedrockModel.
Claude Code + Codex parallel pipeline for bootstrapping Trellis coding specs. CC analyzes the repo with GitNexus (knowledge graph) + ABCoder (AST), creates Trellis task PRDs with full architectural context and MCP tool instructions, then Codex agents run those tasks in parallel to fill spec files. Use when: bootstrapping coding guidelines, setting up Trellis specs, 'bootstrap specs for codex', 'create spec tasks', 'CC + Codex spec pipeline', 'initialize coding guidelines with code intelligence'. Also triggers when user wants to set up GitNexus or ABCoder MCP for multi-agent spec generation.
Multi-baseline counterfactual cost analysis. Compares actual session spend to hypothetical always-haiku / always-sonnet / always-opus routing baselines. Answers "is the routing earning its keep?" Negative savings flag over-escalation; positive savings quantify the router's win.
Per-message cost breakdown within a single session. The drill-down companion to cost-anomaly — when an outlier session is flagged, this surfaces the specific expensive messages so operators can see whether the cost came from output tokens, cache writes, or model escalations.
Use when the user says "/plan-execute", "plan execute", "implement plan", or "execute plan" and provides a finalized plan file path to carry out. Claude orchestrates, Codex writes code, Claude reviews, and Codex fixes issues until the quality bar is met.
Forces exhaustive problem-solving using corporate PUA rhetoric and structured debugging methodology. MUST trigger when: (1) any task has failed 2+ times or you're stuck in a loop tweaking the same approach; (2) you're about to say 'I cannot', suggest the user do something manually, or blame the environment without verifying; (3) you catch yourself being passive — not searching, not reading source, not verifying, just waiting for instructions; (4) user expresses frustration in ANY form: 'try harder', 'stop giving up', 'figure it out', 'why isn't this working', 'again???', '换个方法', '为什么还不行', '你再试试', '加油', '你怎么又失败了', or any similar sentiment even if phrased differently. Also trigger when facing complex multi-step debugging, environment issues, config problems, or deployment failures where giving up early is tempting. Applies to ALL task types: code, config, research, writing, deployment, infrastructure, API integration. Do NOT trigger on first-attempt failures or when a known fix is already executing successfully.
Project orchestrator for the screenwriting skill set (剧本项目主线调度 / 状态存档 / story bible) — a meta-skill that adds no new craft knowledge but routes a screenplay or stage-play project through stages (premise → structure → character → scenes → draft → revision → submission) and a TV / streaming series project (电视剧 / 剧集 / pilot / 一季) through a separate series stage table (engine → character network & season arc → documents/bible → pilot structure → break story & outline → draft → submission), names which sw-* skill to invoke at each stage, defines each stage's deliverable and advisory exit check, and keeps all project state in a story-bible.md file so work can resume across sessions. Use when starting a new script project, resuming one ("continue my screenplay", "where were we"), when the user asks "what should I do next" on a script, when converting a vague idea into a full development pipeline, or when the other sw-* skills are firing individually and the overall process needs sequencing.
Use when context is growing large (50k+ tokens), performance is degrading, instructions are being ignored mid-conversation, or planning multi-agent workflows. Triggers on "lost context", forgotten instructions, or sessions exceeding 30 minutes.
An agent orchestrator for cross-border e-commerce product sourcing. Use this skill when the user needs a complete Temu product selection workflow, five recommended products, or end-to-end product analysis. It automatically calls skills such as amazon-movers-shakers, temu-competitor-search, ali1688-sourcing, and temu-pricing-calculator to produce a complete product selection report meeting the V4.1 standard.
Debug and inspect LLM/AI agent traces using PostHog's MCP tools. Use when the user pastes a trace or session URL (e.g. /ai-observability/traces/<id> or /ai-observability/sessions/<id>), asks to debug a trace, figure out what went wrong, check if an agent used a tool correctly, verify context/files were surfaced, inspect subagent behavior, investigate LLM decisions, or analyze token usage and costs. Also use when raw SQL/HogQL against `events.properties.$ai_input` / `$ai_output_choices` returns empty — message content lives only on the dedicated `posthog.ai_events` table.
Run [implement-spec](../implement-spec/SKILL.md) with Codex handling the implementation passes while you orchestrate, integrate, and review. Use when the user asks to implement a spec with Codex or run implement-spec with Codex as the implementer.
A 6-phase investigation workflow for understanding existing systems. Auto-activates for research tasks. Optimized for exploration and understanding, not implementation. Includes parallel agent deployment for efficient deep dives and automatic knowledge capture to prevent repeat investigations.
Persistent cross-session memory using Memento MCP knowledge graph (mcp__memento__* tools). Recall-before-acting: search memory before starting tasks, on errors, and when receiving corrections. Multi-dimensional search: two queries per recall event (technical topic + process/workflow learnings). Store-after-discovery: persist solutions, conventions, and corrections immediately. Three-step recall: search, open_nodes, traverse relations. WORKING_STATE.md for crash recovery. Self-reminder protocol every 5-10 messages. Activate on task start, errors, corrections, session boundaries, or explicit memory requests. See references/agents-md-setup.md for AGENTS.md integration.
Persistent file-based planning for multi-step AI-agent work. Keeps task_plan.md, findings.md, and progress.md on disk; lifecycle hooks inject selected project planning context. Automatic recovery reads project planning files only. Explicit session-catchup.py --metadata reads same-project local agent session records and emits aggregate counts only; --replay may emit bounded nonce-framed excerpts. Optional gated mode can request continuation only when the host supports it and never runs commands declared in Markdown. The skill has no network upload path. Use for research or work needing 5+ tool calls.
[user/auto] Reviews duplicate, conflicting, or outdated guidance in existing skills, persistent rules, and auto memory. Use it even when behavior following a rule or memory conflicts with user feedback or the latest skill; do not use it for ordinary code errors or creating new skills.
How humans and AI compose in content workflows. Where AI legitimately participates, where humans must own, hybrid workflow patterns, voice ownership preservation, the AI slop problem, disclosure and transparency, team calibration, and the ethics of intellectually honest AI-assisted content production. Triggers on AI content workflow, AI-assisted writing, hybrid content production, AI in editorial, AI slop, AI disclosure, AI usage policy, AI content ethics, voice preservation with AI, team AI calibration. Also triggers when content feels generic despite quality tools, when team AI usage has drifted into inconsistency, or when a regulated or trust-sensitive context requires explicit AI policy.
The primary development workflow for the Archon project (remote-coding-agent). Use this skill instead of any PRP skills when working on Archon code. Routes to 10 specialized cookbooks based on the task: RESEARCH — “how does the orchestrator work?”, “where is session state defined?”, “trace the workflow execution flow”, “what is IWorkflowStore?” INVESTIGATE — “should we use Drizzle or Prisma?”, “what’s the best way to add WebSockets?”, “can we migrate to Turso?”, “how do other projects handle rate limiting?” PRD — “write a PRD for dark mode”, “spec out the notification feature”, “product requirements for webhook retry” PLAN — “plan the auth refactor”, “design the caching layer”, “create an implementation plan for #42” IMPLEMENT — “implement the plan”, “execute .claude/archon/plans/auth.plan.md”, “build the feature from the plan”, “code this up” REVIEW — “review PR #123”, “review my changes”, “code review the diff” DEBUG — “debug the failing test”, “why is streaming broken?”, “root cause analysis on the timeout issue” COMMIT — “commit these changes”, “commit the auth refactor” PR — “create a PR”, “open a pull request for this branch” ISSUE — “report this to gh”, “create a gh issue”, “log it in github”, “file a bug for this”, “create a feature request”. This skill triggers on any development task: researching, investigating, planning, building, reviewing, debugging, committing, or shipping code. Not for running Archon CLI workflows in worktrees (use /archon instead).
Captures knowledge notes and session reports that persist across conversations. Provides full-text search with faceted filters, code structure analysis, and bidirectional cross-referencing between notes and sessions. Use when saving knowledge for future sessions, recalling past work, searching stored context, wrapping up a session, or analyzing code structure. Default: local, zero network. Cloud embedding backends (Gemini, Voyage, OpenAI) are opt-in via `claudemem setup`.
Author and run a durable AI chat agent with chat.agent from @trigger.dev/sdk/ai: the per-turn run loop, why you MUST take streamText from the run argument rather than importing it from ai, returning a StreamTextResult vs calling chat.pipe(), the two server actions (chat.createStartSessionAction + auth.createPublicToken), and wiring useChat to useTriggerChatTransport. Load this when building, modifying, or debugging a chat backend (the agent task or its lifecycle hooks) or its React transport, when declaring typed tools or custom data parts, or when migrating a plain AI SDK streamText route to chat.agent.
Fast, zero-friction capture of technical findings from the current conversation to the wiki's _raw/ staging area. Use this skill when the user says “/wiki-quick-chat-capture,” “quick capture,” “capture this finding,” “save this bug fix,” “capture this gotcha,” “drop this to raw,” “quick save to wiki,” or wants to capture a non-obvious discovery mid-session without a full wiki-ingest run. Writes one _raw/ file per topic cluster in under 60 seconds — no subagents, no QMD updates, no manifest writes. Run /wiki-ingest or /data-ingest later to promote raw files to proper wiki pages.
Designs and deploys custom agent teams for specific business workflows. Interactively discovers business processes, then generates complete team configurations with specialized agent roles, tool access, communication protocols, and handoff rules.
Step-by-step guide for openclaw users to build, deploy, and monetise an AI agent with aixyz. Covers everything from zero: installing Bun, scaffolding an agent, choosing a deployment option, getting a crypto wallet, funding it for on-chain registration, and marketing your agent once it is live.
Diagnoses LLM output failures, including hallucinations, constraint violations, format errors, and reasoning issues. Provides root cause classification, prompt fixes, tool improvements, and new test cases. Use for “debugging AI,” “fixing prompts,” “quality issues,” or “output errors.”
Bootstrap a nao agent for a project: gather warehouse, scope, and extra-context info in one round; look up the warehouse-specific config from nao docs; write nao_config.yaml; run nao init + nao sync; set up the LLM key; and generate the first RULES.md. Use when the user has just decided to use nao on a new project. Only for first-time setup; for editing rules, generating tests, or reviewing an existing context, use write-context-rules / create-context-tests / audit-context.
Diagnose the health of a nao context at any stage of its lifecycle. Use when the user wants a structured review of what's been synced, how RULES.md compares to the target structure, whether every table is documented, whether the data model is MECE, whether tests exist and what their failures reveal, and whether context files are bloated. Outputs a structured audit report with ranked recommendations. Do not use for first-time setup (setup-context) or routine rule writing (write-context-rules).
Use Claude Code as an independent `claude -p` subagent when the user explicitly asks for Claude, wants a second-agent opinion from Claude, or asks to delegate a well-scoped task to Claude. Supports selecting `--model` and thinking/effort level with defaults of the current `opus` alias and `high`.
Analyzes your Claude Code conversation history to identify patterns, common mistakes, and opportunities for workflow improvement. Use when user wants to understand usage patterns, optimize workflow, identify automation opportunities, or check if they're following best practices.
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.